nv50/ir: optimize SUB(a, b) to MOV(a - b)

This helps shaders in UE4 demos, especially with Elemental
(+1% perf). This optimization reduces spilling usage in one
shader which explains the little gain.

GF100/GK104:

total instructions in shared programs :2838551 -> 2838045 (-0.02%)
total gprs used in shared programs    :396706 -> 396684 (-0.01%)
total local used in shared programs   :34432 -> 34416 (-0.05%)

                local        gpr       inst      bytes
    helped           1          19         112         112
      hurt           0           0           0           0

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
index a9172f834b25daa05a1a6ab31e19687a7e2eb9a7..74a5a854e7789ff7dd8f02b931a87988dd639eb6 100644 (file)
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
@@ -576,6 +576,16 @@ ConstantFolding::expr(Instruction *i,
          return;
       }
       break;
+   case OP_SUB:
+      switch (i->dType) {
+      case TYPE_F32: res.data.f32 = a->data.f32 - b->data.f32; break;
+      case TYPE_F64: res.data.f64 = a->data.f64 - b->data.f64; break;
+      case TYPE_S32:
+      case TYPE_U32: res.data.u32 = a->data.u32 - b->data.u32; break;
+      default:
+         return;
+      }
+      break;
    case OP_POW:
       switch (i->dType) {
       case TYPE_F32: res.data.f32 = pow(a->data.f32, b->data.f32); break;